OpenAI updates its new Responses API rapidly with MCP support, GPT-4o native image gen, and more enterprise features

Openai makes Distribute a set of important updates to him New Answers APIAims to facilitate the establishment of intelligent, active agent applications of developers and enterprises.

These accessories include remote support Model context Protocol (MCP) serversCreating an image and integrate code translator tools and search opportunities – Today, the upcoming search capabilities as May 21.

First Started in March 2025Answers API serves as Openai’s tool box for third party developers Deep research and Operator.

From the debut, it has been processed by trillions and supported the trillions and supported the broad use of market research and education to the development and financial analysis.

Popular applications built with API include Zencoder coding agent, Revi’s market scout and MagicSchool’s educational platform.

The main and purpose of the answers API

Answers, in 2025, Openai’s open source agents with SDK, the third party’s initiative to provide access to the same technology, the Openai debuted with the initiative to present its AI agents as a deep research and operator.

In this way, the startup and companies outside Openai can include the same technology for their products and services for their own products and services to use employees for customers and partners.

First, the API, conversation and tools installed for Internet and file search, as well as the use of the computer, as well as the compact orchestra allows you to build autonomous work flows without logic. Openai said that the API had completed the API until mid-2026.

Answers allow you to access the API, model decisions, real-time information and agents, the cause and integration abilities that cause information.

It has made a change to give a single tool set to create a single tool for creating a manufacturer, developers, developers, with minimal friction.

Remote MCP server support expands the integration potential

In addition to this update, it is support for remote MCP servers. Developers can now combine models for external tools and services such as strip, irrigate, stir and Twilio using a few code strings. This allows you to create agents that can interact with the ability, measures and no longer dependent systems. To support this developing ecosystem, the Openai MCP has joined the leading committee.

Answers that increase the update, the agents can do as part of an API call brings newly installed vehicles of API.

Openai’s Hit GPT-4O One option of native image generation model includes a potential useful and quite impressive new features such as real-time flow and multi-turn elegance.

This allows developers to make apps that can produce and edit images in response to user access.

In addition, the code translator is integrated into the API, which meets the models for data analysis, complex math and affiliation.

The tool helps improve model performance in various technical criteria and allows more complex agent behavior.

Improved file search and context processing

File search function is also improved. Developers are now able to search for many vector stores and apply the attribute-based filter to get the most suitable content.

This increases the accuracy of the use of news agents, answering complex questions and increases the ability to operate within large knowledge domains.

New businesses reliability, transparency

Several features are specially designed to meet the business needs. Background mode allows for long-term asynchronous assignments, breaks related to interruptions or network breaks during senses or network breaks.

In subtle summaries, a new addition, the model helps with natural speaking explanations, bears and transparency of the internal thinking process.

Encrypted justification items provide an additional privacy layer for zero data storage customers.

These allow models to recycle the steps before improving both security and efficiency, notifying any information on Openai servers.

The latest opportunities are supported along O-Series models, including Openai’s GPT-4O series, GPT-4.1 series and O3 and O4-mini. These models are now protecting the condition of justification between many vehicle calls and desires, which leads to more accurate answers during a low price and delay.

Yesterday’s price is today’s price!

Despite the expanded feature set, Openai confirmed that prices will remain in accordance with existing rates for new tools and opportunities within the API’s answers.

For example, the code translator is worth $ 0.03 for each session and is included in $ 2,000 per day after the first free gigabyte using the file search.

Web search changes vary according to the size of the model and search context, varies from up to 1000 to $ 50 to $ 50. Creating an image through the GPT-Image-1 instrument, starting at $ 0.011 per image, and is caught in accordance with the level of resolution and quality.

The use of the entire tool is calculated in the significant rates of the selected model without an additional sign for newly added opportunities.

What is the next for answers API?

With these updates, Openai Answers continue to expand what is possible with the API. Developers, enterprises are more integrated, skillful, skilled and reliable AI-chairman applications are accessible to the ready-made features of rich instruments and enterprises.

All features are alive with price and application data through Openai’s documents from May 21.